Semiconductor circuit fabrication methods, as well as other microscopic and nanoscopic manufacturing techniques, have required the development of new imaging techniques, having improved resolution. Improved imaging techniques are also needed in the life sciences. Charged particle beam microscopy, such as electron microscopy and ion microscopy, provides significantly higher resolution and greater depth of focus than optical microscopy. In a scanning electron microscope (SEM), a primary electron beam is focused to a fine spot that scans the surface to be observed. Secondary electrons are emitted from the surface as it is impacted by the primary electron beam. The secondary electrons are detected, and an image is formed, with the brightness at each point on the image being determined by the number of secondary electrons detected when the beam impacts a corresponding spot on the surface. SEMs can also form images from back-scattered electrons as an alternative to secondary electrons. Scanning ion microscopy (SIM) is similar to scanning electron microscopy, but an ion beam is used to scan the surface and eject the secondary electrons. Ion microscopes are also able to form images using secondary ions.
In a transmission electron microscope (TEM), a broad electron beam impacts the sample and electrons that are transmitted through the sample are focused to form an image of the sample. The sample must be sufficiently thin to allow many of the electrons in the primary beam to travel though the sample and exit on the opposite site. Samples are typically less than 200 nm thick and often much thinner.
In a scanning transmission electron microscope (STEM), a primary electron beam is focused to a fine spot, and the spot is scanned across the sample surface. Electrons that are transmitted through the work piece are collected by an electron detector on the far side of the sample, and the intensity of each point on the image corresponds to the number of electrons collected as the primary beam impacts a corresponding point on the surface. The term “TEM” sample as used herein refers to a sample for either a TEM or an STEM and references to preparing a sample for a TEM are to be understood to also include preparing a sample for viewing on an STEM.
Because TEM samples are very thin, preparation of the samples can be delicate, time consuming work. Thickness variations in thin samples, particularly samples less than 100 nm thick, can result in sample bending, over-milling, or other catastrophic defects. The preparation technique determines the quality of structural characterization and the ability analysis of the smallest and most critical structures.
TEM samples can be broadly classified as “cross-sectional view” samples or “planar view” samples, depending on how the sample was oriented on the work piece. If the face of the sample to be observed was parallel to the surface of the work piece, the sample is referred to as a “planar view” or “plan view” sample. If the face to be observed was perpendicular to the work piece surface, the sample is referred to as a “cross-sectional view” sample.
Several techniques are known for preparing TEM samples. Some techniques may involve cleaving, chemical polishing, mechanical polishing, or broad beam low energy ion milling. Combinations of these techniques are also possible. These methods are disadvantageous because they often require that the starting material be sectioned into smaller and smaller pieces, thereby destroying much of the original sample. They are also generally not site specific.
Other techniques generally referred to as “lift-out” procedures use a focused ion beam (FIB) to cut the sample from a substrate while greatly limiting or eliminating damage to surrounding areas of the substrate. These techniques are useful for analyzing the results of semiconductor manufacture, for example. Using lift-out techniques, a sample can be analyzed in any orientation, such as a cross section or plan view. Some techniques extract a sample that is sufficiently thin for use in a TEM without additional preparation. In other techniques the extracted sample is thinned further before observation.
Techniques in which the sample is extracted from the substrate and moved to a sample holder within the FIB vacuum chamber are referred to as “in-situ” techniques. Techniques in which the sample is removed from the work piece and moved to a sample holder after the work piece is removed from the vacuum chamber are referred to as “ex-situ” techniques.
A cross-sectional sample is typically prepared from a larger bulk sample by milling away material with an ion beam to create trenches on either side of the region of interest, leaving a thin section referred to as a “lamella.” The lamella is partly severed from the sample substrate by ion beam milling around the bottom and the sides of the lamella until it is connected to the substrate only by a small amount of material. In some cases the connecting material might be “tabs” on either side. In an in-situ process, a sample manipulation probe is then brought in close proximity to the thin sample. The probe is attached to the thin sample, typically by beam-induced deposition of a material from a precursor gas, but other methods can be used, such as electrostatic attachment. Beam deposition can be done with either the FIB or SEM. The material connecting the thin sample to the work piece is then milled away (or mechanically broken) to leave the sample connected only to the manipulation probe. The probe, with the sample attached, can then be moved to a different position where the sample can be attached to a TEM sample holder, called a “grid.” The probe is brought very close to a selected part of the TEM grid, and the lamella is attached to the grid, typically by beam-induced deposition. Once the sample has been attached to the grid, the sample probe can be disconnected from the sample, for example, by severing connection with the FIB or merely by moving the probe and/or sample to break the connection. The lamella may be processed further after attachment to the grid.
The process of creating and extracting a lamella and transferring it to the sample grid is a delicate and time-consuming procedure, often requiring about 45 to 90 minutes per sample, and requiring the constant operator attention of a skilled operator. For total analysis of an area of interest on a semiconductor wafer, it may be desirable to analyze as many as, for example, 15 to 50 or more TEM samples. When so many samples must be extracted and measured, the total time to process the samples from one area can be hours or even days. Thus, even though the information that can be gained through TEM analysis can be very valuable, the process has been prohibitively time consuming for manufacturing process control and other routine procedures.
Improving the speed at which lamellae can be prepared for imaging therefore would provide significant advantages in both time and potential revenue by allowing work pieces selected for analysis to return to the production line more quickly. Automation of the lamella preparation process would not only speed up the process but also reduce the level of expertise required for operators, representing an advantage for the manufacturer. In addition, a skilled operator may be performing other tasks while automatic operations are being performed, increasing the throughput of the procedure.
Due to the precision required to mill, extract, transfer, and deposit a lamella on a sample grid, the process has not adapted itself to automation. As the lamella thickness is reduced, it becomes more likely that the region of interest will be excluded from the lamella. Lamellae under 100 nm in thickness, particularly lamellae under 70 nm, are difficult to produce either manually or automatically.
As lamellae becomes thinner, they can warp due to thermal or mechanical stress, changing their positions relative to the beam, which can ruin the lamella by allowing the beam to impact the region of interest. Thickness variations in the lamella can result in sample bending, over-milling, or other catastrophic defects that render the sample useless. In addition, the sample probe for manipulation of the lamella must be placed with extreme precision when preparing to extract the lamella from the substrate, and also when landing the lamella on the sample grid. These factors combine to make the preparation of lamella for analysis an exceedingly difficult process to automate.